Re: [DCP-o-matic] Question about loudness

Uhmmm….my apologies to the thread…I didn’t mean for my rambling document on Loudness to get sent in that condition…you got to see the insanity of my writing process. I did give an overview presentation about Loudness a couple years ago, which was Part I: Loudness In Cinema <https://www.dcinematools.com/index.php?option=com_content&view=article&id=1879:loudness-in-cinema-ibc-2016-presentation&catid=114:white-papers-exhib&Itemid=176> …I am still writing Part II which deals with the measurement techniques, the current science and looking at routes to the future. Suffice to say that Carsten is correct. We don’t know enough yet. The attempts that the EBU and ATSC separately made to achieve the LUFS and LKFS metrics are very broadcast specific – and rely in the end on the consumer having a remote control in their hand to set the nominal comfortable level for their situation – and to some degree are meant to get around the ability for clever users and new compression technology to game the LEQm metric (which is the metric that pre-show trailers was designed to follow) in a system that uses much less dynamic range than movies. But let’s be very clear. None of those, LEQm, or LUFS or LKFS, are appropriate for long form products that use a great deal of dynamic range…which are just two of the important factors that separate cinema from broadcast. The current proposals in the air are to add a 10 minute moving window for the measurement to even up the numbers, which is called LEQm10 or LUFS10 or LKFS10…but there are other things missing too. Dialog gating gets mentioned a lot, which was just added to LUFS version 3. …though, for some reason, Netflix’s just released spec for ads made for their system says LUFS v1 with dialog gating. There are other places to get dialog gating, but Dolby is now giving their version for the asking and the presumption is that this will be politically more acceptable than others. All this to say, we are a long road away from getting to the path that will take us to the result. Outside of not having a real metric, and outside of the stretched resources of a small group of volunteer engineers who could develop and test it – much much smaller than the broadcast world which still took years – and outside of the politics involved between creatives that don’t want to be told what to do with their art, and outside of the politics involved with exhibitors who don’t want to be told what to do with their customers, and outside of the fact that there is currently no place to put the number in the DCP package that might influence the automation systems…well, yeah, research, development and productization might take a few extra weeks beyond “it takes what it takes" before logic and loudness tranquility is reached. PS – the leq-nrt code that Luca spent years refining, is available on GitHub for use. But it is a long trek to get it into ffmpeg, and yet again, it takes people who volunteer the time to make the case and support the effort and make certain it is done correctly. So, volunteers anyone? (The good part is that I understand that LEQm10 is already incorporated.) Again, apologies for going long. C J Flynn